tracksnap

Stateful RSS/Atom feed tracker. Remembers what you've seen — shows only what's new.

pip install tracksnap

The idea

Most feed readers show you everything, every time. tracksnap is different: it tracks which items you've already seen in a local SQLite database and only returns the delta. Run it in a cron job, pipe it into another tool, or use it as part of an agent workflow — you only ever process genuinely new content.

Usage

# First run: shows all items, marks them as seen
tracksnap https://hnrss.org/frontpage

# Second run: shows ONLY items you haven't seen yet
tracksnap https://hnrss.org/frontpage

# Show everything regardless of history
tracksnap https://hnrss.org/frontpage --all

# Forget history for this feed, then show all items fresh
tracksnap https://hnrss.org/frontpage --reset

# Limit number of items
tracksnap https://hnrss.org/frontpage --limit 5

# Output formats
tracksnap https://hnrss.org/frontpage --output json
tracksnap https://hnrss.org/frontpage --output table
tracksnap https://hnrss.org/frontpage --output csv

State storage

History is stored in ~/.local/share/tracksnap/state.db (SQLite). Each feed URL gets its own set of seen item IDs. You can override the path with the TRACKSNAP_DB environment variable.

Automation / agent use

# In a cron job — only process genuinely new items
tracksnap https://example.com/feed.xml --output json | jq '.items[].title'

# Feed into briefsnap or another pipeline tool
tracksnap https://example.com/feed.xml --output csv >> new_items.csv

# Inspect tool capabilities (ACLI-compliant)
tracksnap introspect
tracksnap skill

Feed support

• RSS 2.0 — standard <item> elements with guid, link, title, author, pubDate, description
• Atom — <entry> elements with id, title, link, updated/published, author, summary/content

Part of the snap family

tracksnap is the stateful companion to feedsnap (stateless RSS reader) and fits naturally into workflows with briefsnap (morning digest aggregator).

License

MIT